Help:Wikipedia: The Missing Manual/Building a stronger encyclopedia/Deleting existing articles

One of Wikipedia's strengths is how easily editors create new Wikipedia articles—quick, efficient, no bureaucracy to get permission from. But some editors abuse the privilege, creating nonsense articles, attack articles, or promotional articles, for instance. Others just make mistakes because they don't understand Wikipedia's rules and purpose: They create an article about a non-notable topic, or a subject already covered by an article of a different name, or containing a definition (which should be a Wiktionary entry)—not an encyclopedic topic, for example.

In order to keep Wikipedia the useful encyclopedia that it is, articles with all these problems need to be improved or removed. For Wikipedia, the challenge is to have processes that delete inappropriate articles (more than a thousand a day) while keeping potentially good articles and avoiding offending well-intentioned editors. In this chapter, you'll learn the different ways to deal with problem articles (often without resorting to one of the three deletion processes). You'll also learn what recourse you and other editors have if you feel that an article was deleted inappropriately.

Responding to problem articles
You can stumble across a problem article in a number of ways—doing a search, reviewing the User Contributions page of an editor, looking through Special:Newpages, or just clicking "Random article" to see what you see. When your first reaction is, "You're kidding!" or "I can't believe this is a Wikipedia article," your second reaction should be to analyze the article, systematically. There may be grounds for a quick deletion, or you may have to do some further research.

First analysis
Here's a systematic approach to dealing with an article that you think may not belong in Wikipedia. Ask yourself the following questions in order:


 * Is this vandalism? What you see may be the result of someone deliberately damaging a perfectly good article. You should look at the article history (as discussed in detail in Chapter 5: Who did what: Page histories and reverting) to see what the article looked like in previous versions.


 * Obviously, any major decrease in the size of the article is grounds for suspicion. If the article was created recently, check the similarity of the first version to the current one. If it's been around for a while, then check the versions from 3 and 6 months ago. If you do find vandalism, revert the article to its last good version.


 * Is this article worth trying to salvage? Some articles are just junk, not worth the time it would take to fix them. Wikipedia has a Criteria for Speedy Deletion policy (shortcut: WP:CSD). The candidates for immediate deletion include patent nonsense and gibberish; test pages ("Can I really create an article?"); attack pages, including entirely negative, entirely unsourced biographical articles (don't attempt to salvage these); no evidence that the subject is notable enough to belong in an encyclopedia; and no content (only links or chat-like comments about the subject). If the article is any of these things, then nominate it for speedy deletion, as discussed on the section about Speedy Deletion.


 * Does the article look like blatant advertising, or does it appear to be copied and pasted from a Web site (blatant copyright violation)? These criteria both call for speedy deletion, though for the second you'll need to find the source of the text before nominating the article for speedy deletion. But also ask yourself: Is there something salvageable here, something worth an article? If you think there might be, read the next section.

Notability and verifiability
At Wikipedia, notability (see the section about notability) is an important criterion for creating a new article. Since Wikipedia contains no original research, notability is determined by coverage of the topic in upstanding, independent sources, like books, newspapers, and sometimes Web sites. (For full details, check out the guideline Notability; shortcut: WP:N.)

At its heart, deleting an article—other than by speedy deletion—is the Wikipedia community's way of saying that reliable sources of information simply don't exist out there in the world to make the article lengthy enough to stand on its own. Of course, it's impossible for you, or other editors, individually or collectively, to know if that's really true. You simply have to make your best guess, because it's impossible to prove something's non-existence.

The challenge of notability and acceptable sources
Consider, for example, an article about a high school teacher, describing her classroom manner, the names of her pets, and how much her students like her. Is that ever going to be acceptable as an article? Is it likely that a number of newspaper articles have been solely about this teacher? While it's possible (if, say, the article's about the National Teacher of the Year), it's also unlikely. If a quick search turns up nothing, then the article should be tossed.

Compare, by contrast, an article about a 19th-century Portuguese poet. If the article makes no claims about notability (says nothing about "famous", or "well-known", then it's a candidate for speedy deletion. But you're probably not an expert on 19th century poets, of any nationality, so for you it's probably a sheer guess as to whether you're looking at a vanity—someone's great-great-great-grandfather—or someone once famous who's faded into obscurity, but whose life and accomplishments are well-documented in books published more than a hundred years ago.

Possible responses
You have three options when faced with a problematical article that you decide you don't want to nominate for a speedy deletion until you evaluate notability:


 * If you're inclined toward non-notability, do a quick search or two to see if you're mistaken. Spend, say, 3 minutes on the matter, and then take a look at the alternatives to deletion in the next section.
 * If you're inclined toward notability, but don't have time (or inclination) to spend much time fixing the article, still do a quick search or two (again, just a couple of minutes), add anything you find to the article (as an external link), and then add a cleanup message box to the article if it's lacking one. (You'll find nicely organized ones at the page Template index, shortcut WP:TM.)
 * Finally, and best of all, if you have the time, you can fix the article. If the topic is notable but the article is missing sources, put in some citations (the section about Citing Sources in Chapter 2: Documenting your sources). Chapter 18 also discusses a general approach for improving articles, though it's tailored more for long articles than for stubby ones without any sources.

Alternatives to deletion
If you find a problem with an article, consider some alternatives that might make the author of the article feel better, while protecting the quality of Wikipedia. These alternatives also have the advantage of not requiring an administrator's help:


 * If there's nothing on the page worth keeping, try to identify a related page that's useful to readers, and create a redirect (the section about Redirects). Redirects work particularly well when the editor who created the page has lost interest and stopped editing, and that would be the only person objecting.
 * Move the page to a new subpage in the user space of the editor who created it. This approach (called userfication) is particularly good for articles where it's unclear that notability can ever be established. It challenges the editor to find sources, or let the page languish. Make sure that you put a note on the user talk page about what you've done, with a link to the new subpage, and offer to help explain policies and guidelines further if the editor would like. (For further information, see Userfication, shortcut WP:UFY.)


 * Where there's a bit of useful information, but only a smidgen, and you have doubts that the article can be easily expanded (or ever expanded into a real article), merge the information into another article on a broader topic, and put a redirect in place. In the best of worlds, you can create a new section in the article on the broader topic, and link to that. In any case, try to link to a section of the article, to encourage the editor to expand that.


 * Here too, leave a note on the user talk page for the author of the article, saying that you've merged the information from that article into another article, and that you hope the editor can expand what's in that larger topic.


 * You can also copy the information to a sister project such as Wiktionary (most commonly), or at least put a template on it suggesting that it be copied, and hope that someone else will do so. Once the information is copied across, the page can be made into a redirect, or deleted.


 * Obviously, this final option is limited to things that really do belong elsewhere, such as a dictionary definition, and only when such an entry doesn't exist at all in the other wiki. (If it does exist, then copying the article in its entirety isn't really an option.) But if you can move the material, it may make the editors feel that they've contributed to something, even if not Wikipedia. (For more information, see the Meta page Help:Transwiki and the guideline Wikimedia sister projects; shortcut: WP:SIS.)

Three ways to delete an article
Only administrators can delete articles outright. Your job is to ask for deletion using one of three methods:


 * Speedy (CSD). If the article meets any of the criteria in the Criteria for Speedy Deletion (CSD) policy, then go for it.
 * Proposed deletion (prod). If the article doesn't quite qualify for a speedy deletion, then use the Proposed Deletion (prod) template, which starts a 7-day countdown.
 * Articles for Deletion (AfD). If you're in doubt, start a discussion at the Articles for Deletion page about whether or not the article is worth keeping.

Each of the three methods follows a different procedure, discussed next.



Speedy deletion
Speedy deletions are based on specific criteria, listed in the policy Criteria for speedy deletion (shortcut: WP:CSD). (See Figure 19-2.)



If you want to propose a speedy deletion for an article, you need to cite a specific criteria found on this page, either from the "G" (general) series (which applies to all types of pages, not just articles), or the "A" (article) series. For example, criteria G2 applies to test page (pages created by a new editor, just exploring possibilities).

You use the criteria to determine the template to post on the article; once you've done that, an admin shows up fairly quickly, reviews your nomination for deletion, and decides whether to remove the template (as an error) or accept it and delete the article. If you've made a particularly egregious error in placing the template on the page, you'll also probably get a note to that effect from the admin.

Initial review
Take, for example, an article called Timberwilde Elementary School. The first step in any speedy deletion is to figure out whether the article in fact meets any specific Criteria for Speedy Deletion (CSD). The entire article under consideration in this example reads, "Timberwilde Elementary School, Built In 1980, Is A School In The Texas District Northside Independent School District." That's better than the original posting, which read, in its entirety, "Timberwilde Elementary School Is The School Of Awesomeness! It Has A Graet Varitey Of Students Teachers And Staff! Visit Us At NISD.NET/TIMBERWILDE Click On Campus Webpage."

This article looks like it fits CSD number A7, "No indication of importance/significance." But before you pull the trigger, you should check five other things:


 * The article talk page. In this case, it has a template for WikiProject Texas, and the history of the page shows no postings relating to the contents of the article. So, there's no indication of importance/significance. (The article being marked as within the scope of a WikiProject does not prove notability — in this case, it's because the article included the word "Texas.")
 * The article history page. In this case, there were basically two editors—one who created the article (and never came back to expand it) and one who did a tiny bit of cleanup. That's further evidence of non-notability. If the school were notable, a bunch of editors would have contributed at least a little bit each (assuming an article isn't brand-new).
 * The What links here special page. Click the link in the left margin to see it. In this case, no other articles link to this article.
 * The User contributions page for the editor who created the article. In this case, it shows only two edits, both on the same day (more than a month ago). The second edit was of a regular article. So there really isn't any reason to assume that the author of the article understands Wikipedia rules.
 * A Google or Yahoo or other search of the topic. In this case it shows nothing significant.

So, from the review, it seems fair to say that this article looks like an A7.

Making a Criteria for Speedy Deletion (CSD) nomination
Once you've completed your initial review and found a criterion for deletion that fits, you can turn to the second step: actually nominating the article for deletion. Here's the process:

1. Further down the CSD page, in the "Deletion templates" section, look for the templates that match the specific criteria you've found.


 * In this case, the choices are db-bio, db-band, db-club, db-corp, db-group, and db-web. All rather specific, and none quite on point. The topic here is an organization. That's OK under criteria A7, but there isn't a specific template for it.

2. When the specific templates for a criterion don't fit the situation exactly, use a general criterion.


 * You'll find them at the top of the "Deletion templates" section, as shown in Figure 19-3. In this case, a good template is .

3. Edit the article page, adding the template at the top of the edit box.


 * Paste it all the way at the very beginning of the wikitext, as shown in Figure 19-4.

4. Add an edit summary, being sure to mention both "speedy delete" and the specific CSD (in this case, "A7"). Preview the page, and then press 'Publish changes'.


 * You see your request for deletion in a message box at the top of the article, as in Figure 19-5.

5. Post a notice on the user talk page of the editor who created the article.


 * Doing so isn't just being nice (although it is nice). It also creates a record of the fact that this editor created this article. If the article does get deleted, the edit that created it will no longer be visible on the editor's User contribution page, so this user talk page posting will be the only notice to other editors, useful if the pattern persists.

Once you complete these steps, you're done for now. Check back in a day or so. What you do next depends on what's happened to the article.


 * Most of the time the article's gone, deleted by an admin.
 * If the admin reviewing the action rejected the nomination, then you usually want to move to the next step—deletion via proposed deletion (nicknamed prod), as discussed in the next section.


 * Maybe the reason for rejection is a persuasive argument against any type of deletion. If so, then it's time to move on to something else. But usually rejection is because the admin decided that the CSD criterion really didn't apply. That's fine—there's more than one way to get an article deleted.


 * If the editor who created the article deleted the CSD template (which is against the rules, but it happens), put it back up. You can just revert to your version. Also, post a warning on that editor's talk page about violating the rules.
 * If another editor (besides the one who created the article) deleted the CSD template, look at the stated reason. Hopefully there's one in the edit summary. Try to discuss the matter with that other editor, with the goal of arriving at consensus (you both agree whether the article should stay or go). Or try to get a commitment by the other editor to fix the article by a certain date, or let it get deleted.


 * If you can't reach a consensus or find a way to fix the article, then go straight to the third type of deletion, an Article for Deletion  (AfD) (the section about Articles for Deletion).

Proposed deletions (prod or: PROD)
Proposed deletion (shortcut: WP:PROD) is a way to nominate an article for deletion, used when you think the deletion won't be controversial. If the deletion meets the more stringent criteria for speedy deletion (the section about speedy deletion), then use that. But often you can't; for example, when an article contains a bit of reasonable content, or indicates that the topic is notable.

Nominating an article as a proposed deletion starts a 7-day clock. If no other editor objects, then an administrator shows up after 7 days, reviews the nomination, and, if it looks okay, deletes the article.

When you can't use the proposed deletion process
As convenient as the proposed deletion process it, you can't use it on the article if any of the following are true:


 * The article has previously been proposed for deletion. If one editor puts a prod template on an article and another editor removes it (and the removal isn't part of a vandalizing edit), then the proposed deletion is contested.
 * The article has previously been undeleted. To be undeleted, an editor has to make a case that the article was deleted by mistake. That makes deleting it again at least somewhat controversial.
 * The article has been previously nominated for deletion using the AfD process. Presumably, the editor wouldn't have gone to AfD unless deletion at that time was considered potentially controversial. So deletion may still be controversial.

You can check for these three situations by looking at the article history. Normally it's not very long; if it is, you should rethink the prod (consider an AfD instead).

If you can't do a proposed deletion because of one of these three circumstances, your choices are to drop the matter or use the AfD process (the section about Articles for Deletion) to nominate the article for deletion.

Initial review
Take as an example the article SQL-I. SQL-I, the article says, is a programming language, "a tool anyone with basic knowledge of SQL syntax can learn in one day." It "provides system administrators, advanced users and independent developers the option to write their own plug-ins." You could argue that this article's a candidate for speedy deletion, either as an A7 (no claim to notability) or a G11 (blatant advertising). But there's a bit of meat to the article (the total text is about three times what's quoted here), and a prod is just as efficient as a CSD (it just takes longer), so you decide to do a prod instead.

When you're considering a proposed deletion, first you appropriately research the article. Researching is critical because an incorrect prod is worse than an incorrect CSD. Articles that meet the CSD usually don't contain much useful material, so if the deletion's in error, very little is lost. With a prod, there's normally more information in the article—more of a potential foundation for other editors—as in this case. So if it's deleted by mistake, more is lost.

Since SQL-I is a current software product, a Google or Yahoo search seems reasonable. In Google, searching for both the product name and the company that sells it, and restricting results to English, yields a total of seven results, none of which are of citable quality. Apparently, SQL-I is not notable.

Next, you want to look at the same four other things that you do for potential CSDs:


 * The article talk page. In this case, it doesn't exist, so check it off the list.


 * If there is a talk page, see if it has anything helpful, like some suggested sources. You also want to see if there's any indication that the article's been through the AfD process (the section about Articles for Deletion); if so, you can't do a proposed deletion. (You can still nominate the article via the AfD process again if you want.)


 * The article history page. In the SQL-I case, one editor created the article, made six more edits on the same day (almost a year previously), and never came back. There are no other contributors. A bot tagged the article as uncategorized, and another editor added a category, but that doesn't count as people actually interested enough in the article to come across it on their own and expand it. So the history page has further evidence of non-notability. Also, there's no indication anyone's ever proposed deletion.
 * The What links here special page. In this case, one other article links to this one. And, as you'll find out next, it turns out that other article was created by the same editor.
 * The User contributions page for the editor who created the article. Reviewing this page is the most important step before doing a prod. If the editor turns out to be an active contributor to a variety of articles, then you want to ask that editor about the article, not nominate it for deletion.


 * In this case, the editor created three articles—this one, one about another product of the same company (this product seems a bit more notable, though the article is similarly unsourced), and a third about a mountain resort (which seems the most notable article of all). There are no other contributions, and the last edit was more than 10 months ago. So, there's no compelling reason to start a dialog before proposing a deletion—it's likely that no one's home.

Making a "PROD" nomination
Once you've completed the initial review, move on to the second step: actually nominating the article for deletion. Here's the process:

1. Edit the article, placing the  template at the top, and then change the word "reason" to explain your rationale.


 * Place it at the very beginning of the wikitext, as shown in Figure 19-6.



2. Add an edit summary, being sure to mention proposed delete.


 * If you want to add some information on your reasoning, that might save an editor time going to the article to see if the "prod" is justified.

3. Preview the page, and then hit 'Publish changes'.


 * You see a message box asking for deletion, like the one in Figure 19-7.

4. Post a notice on the user talk page of the editor who created the article, and anyone else who was a major contributor.


 * The notification is more than a courtesy. Ideally those editors will come back and fix the article before it's deleted. On the other hand, if you're notifying more than two or three people, you're either over-notifying or you shouldn't have proposed the article for deletion in the first place.

After you're done, check back every day or two. What you do next depends on what other editors do:


 * If no one removes the prod template, then in 7 or so days an admin will probably delete the article.
 * If the admin reviewing the proposed deletion rejected it, then consider the next (and final) step—nominating the article for deletion via Articles for Deletion (AfD), as discussed in the next section.
 * The reason for rejection might be a persuasive argument against any type of deletion. If so, then you're done here. But it's more likely that you missed one of the three reasons why prodding wasn't allowed (see the section about proposed deletions). Fortunately, you can still turn to the AfD process.
 * If another editor has removed the prod template, they should have explained why. But even if they didn't, you should still consider the deletion to be contested. If that editor or others get busy improving the article, then you need to reassess your intent to get the article deleted. If nothing has changed, however, turn to the AfD process. Don't repost the prod template after another editor removes it; removal alone is considered enough to make the deletion controversial.

Articles for Deletion (AfD)
When you nominate an article for deletion in the AfD process, you must have a good reason. You're asking other editors to spend their time reviewing the article and commenting on the reasons given for the nomination, so don't waste their time. The CSD and proposed deletion processes are preferable, but they don't apply to some articles. Before initiating AfD, explore alternatives to deletion (the section about alternatives to deletion). AfD is a last resort, when an article is unsalvageable and there are no alternatives. Articles for deletion (WP:AFD) lists measures you should consider first, like improving the article or making it a redirect (Figure 19-8).



Justification for an AfD nomination
When you nominate an article for deletion using the AfD process, you get to a point in the process where you have to give a reason. Don't make the mistake of reaching that point only to realize that you don't actually have a good reason. Or worse, don't make the mistake of completing the AfD process and discovering, when discussion starts, that what you thought was an acceptable reason was not.

Reading two policies will keep you from making such mistakes:


 * The section "Before nominating an AfD", at WP:AFD, is a quick, step-by-step summary of everything from alternatives (tagging for cleanup, doing a redirect, and so on) to guidelines you should know (WP:BIO, WP:CORP, and so on) to some suggestions on the technical aspects of an AfD nomination.
 * The section "Reasons for deletion" at Deletion policy (shortcut: WP:DEL) is a list of 15 reasons (some not pertaining to articles) that probably cover almost all successful AfDs.

You need just a couple of minutes to read each of these. Once you read them, you're well-prepared to do an AfD nomination if that is, in fact, appropriate. You may have to follow some links and do some more reading. If so, what you read will be useful in your normal editing work as well, so consider it an investment, not drudgery.

Nominating an article at AfD
The following steps walk you through the AfD process using the article Salmon fishing with the Dry Fly as an example. An editor had proposed the article for deletion (with a "prod" template, as discussed above). The editor who created the article removed that template, without any explanation.

This example assumes that you've already tried, unproductively, to discuss with the author the unacceptability of such a "how-to" article. You also did a redirect to the article Fly fishing, but it was reverted, leaving no choice but AfD.

1. Review the "Reasons for deletion" section of Deletion policy (shortcut: WP:DEL) and prepare your argument as to why the article should be reviewed.


 * If you can't point to one of the reasons here in that policy, like "All attempts to find reliable sources in which article information can be verified have failed", it's more than likely that the nomination won't result in a "delete" decision.

2. Open the article for editing. Add the template {{subst:afd1}} at the top of the text in the edit box, and add the recommended text to your edit summary, changing "PageName" to the article's actual name.


 * Preview the page; if you see an article message box that starts, "This article is being considered for deletion ...", as shown in Figure 19-9, then publish the edit. (Otherwise, fix the template.)

3. In the AfD message box at the top of the article, click the "Preloaded debate" link (it's in small print, in the second grouping of text).


 * That link takes you to a page with a five-step set of instructions (Figure 19-10). These five steps take you through the process of creating a discussion page for this specific AfD, and adding that page to the daily log of AfDs so that other editors can find the page. Each is discussed below.

4. Do the first two steps—select the article's category, and add a reason to the standard template.


 * The edit box for the page should look like Figure 19-11.



5. Copy (Ctrl-C or ⌘-C) the text listed in step 3 of the instructions (in this case, ). Then open a new tab or window for the link "today's AfD log", which is also in step 3 of the instructions.


 * You arrive at the log page for the current date, in edit mode. In the edit box, scroll down until you see the place where you want to paste this text (see Figure 19-12).

6. Flip back to the window where you were editing the discussion page, as shown in Figure 19-11 (this is the page with the five steps on it). Copy the text in step 4 to the clipboard. Now flip back to the log page (Figure 19-12), paste this text into the edit summary, and then publish the page.


 * Don't bother with a preview, or with looking at the log page after you've published your edit—until you finish creating the discussion page, the log page doesn't show the discussion page correctly. Next, you'll finish the discussion page.

7. Tab to the edit summary box. The box should already be filled in. If not, paste (again) the text from step 4 into that edit summary box. Turn on the "Watch this page" checkbox, if you use your watchlist regularly, since you'll want to watch the discussion. Do a preview, and then publish the page.


 * What you see should be similar to Figure 19-13. You're almost done.

8. Go back to the log page and check that it shows your discussion page (it should be at the top). If it does, close it.


 * If there's a problem on the log page, then fix it. Looking at the wikitext, figure out what's different about your listing, and change that.


 * The final step is to notify the editor who created the article that it's being considered for deletion.

9. Go to the discussion page, and then click the link to the article. Copy the small print at the very bottom of the message box (in this case,   and paste it to a new section on that editor's user talk page.


 * Leave the "Subject/headline" box blank—the template will take care of that.

You're done—except for the actual discussion, of course.

Participating in AfD discussions
An AfD discussion normally runs for seven days, though it can be closed earlier if the discussion is clearly, massively one-sided. (The page Snowball clause, shortcut: WP:SNOW, explains the concept of not letting a process continue when continuing makes no sense. The guideline Speedy keep, shortcut WP:SK, explains when and how to close an AfD as a "keep" before the full seven days are up.)

If you've nominated an article for deletion, be judicious in adding comments to the discussion. You should have made your point when you gave the reason for the nomination. If you keep repeating that, or variants of it, or argumentatively question the reasoning of editors who want to keep the article, you'll lose credibility.

As nominator, think of your role as a facilitator. You've created the agenda, now let the Wikipedia community decide. If you have additional information to offer in response to a posting of another edit, then provide it. Otherwise, let the conversation flow. It's not a win or lose competition; it's a discussion about making Wikipedia a better encyclopedia.

If you want to participate in AfD discussions you didn't start, here are some suggestions:


 * Start by participating in discussion about topics that you know. Use the "Categorized discussions" section at WP:AFD to narrow the range of articles on which you focus. As you get more familiar with AfDs, you can contribute to discussions about topics you know less about, because you'll know more about relevant policies and guidelines.
 * Read the essay Arguments to avoid in deletion discussions (shortcut: WP:AADD), which is a collection of unpersuasive arguments that experienced AfDers have heard over and over: "Delete as unencyclopedic", "Keep—it's clearly notable", "Delete because WP:RELEVANT is only a guideline"; "Keep because I like it", and so on.
 * The best way to win the argument for keeping an article is to improve it with good citations, and text from good sources, while the discussion's going on. If you can fix the problems that lead to the deletion nomination, even the editor proposing the deletion should be happy to see the article kept.

After an article is deleted
Someday, an article you created or worked on may be deleted, and you won't know or won't agree with the reason for the deletion. If so, remember that very little is actually deleted in Wikipedia—it's still visible to admins, just not to regular readers. And Wikipedia's processes are not infallible, so you have some options.

First, read Why was the page I created deleted? (shortcut: WP:WWMPD). Among other things, this page explains how to find out why an article was deleted. If an article was deleted as a result of a proposed deletion, any administrator will usually restore it upon reasonable request. Follow the link at WP:WWMPD to post such a request.

A second option is to try to persuade the admin who made the deletion that it was in error. This option is worth trying only for CSD deletions, since prods can be reversed on request, and AfDs won't be reversed just because you ask nicely. Before you make such a request, do your homework—does the CSD deletion really seem unreasonable? Don't, for example, argue that the deletion was wrong because the article could have been fixed. CSDs are based on what was actually in the article, not the article's potential.

Another option is simply to start over. If the article was short, you've really not lost much if it was deleted. (You might even find a copy at Google—click the Cached link rather than the main link.) Just be sure that you start the article in your own user space, and don't move it to mainspace (where all regular articles are) until you're sure it can survive on its own. CSD criteriom G4 allows the speedy deletion of a "substantially identical" copy of any article deleted via AfD, particularly where the problems identified in the AfD discussion have not been addressed.

Sometimes you might acknowledge that the deletion was probably right, given the shape the article was in, but you think you could fix it. If so, it would help to have a copy of what was deleted. Check the page Category:Wikipedia administrators who will provide copies of deleted articles, and make your case to one of the admins (check their User contributions page first, to make sure they're still active). Ask that a copy of the article be put into your user space, as a subpage. If the article wasn't libelous, a massive copyright infringement, or an attack page, you're likely to succeed.

Finally, you can initiate a deletion review process, at the page Deletion review (shortcut: WP:DRV). This page is for appeals to restore deleted pages (and also for appeals to delete pages which were closed as "keep'" in an AfD discussion). Before you do so, read the section, "What is this page for?", which explains that DRV is for cases where you believe the process was wrong, or where "significant new information has come to light." DRV is not a place to say that you didn't like the outcome, or a place to go in the hopes that a new discussion can occur that'll lead to a different result.